Predictive control: propositions for the design methodology

Abstract

Deals with predictive control tuning. The study is carried out by simulating second order linear or non-linear processes, corresponding to simulator models of industrial furnaces. The interest for predictive control is tied to its attractive concept and properties, as well as to its widespread use in industry. Unfortunately many parameters have to be tuned in order to achieve efficient control laws. The goal of the present work was to concentrate on the model-based predictive controller (MPC) tuning methodology in order to derive a nearly automatic design strategy. Firstly, new design parameters are introduced in order to improve the results when the reference signal is known in advance. Secondly, propositions are made for a heuristic tuning method based on expert rules. A nonlinear furnace model is used to illustrate the guidelines.